Role of regulatory T cells and associated
chemokines in human gynecological tumors
Doctoral Dissertation of: Christoph Paul Freier
Direktor: Prof. Dr. med. Sven Mahner
Role of regulatory T cells and associated chemokines in human
gynecological tumors
Dissertation
zum Erwerb des Doktorgrades der Naturwissenschaften an der Medizinischen Fakultät der Ludwig-Maximilians-Universität zu München
vorgelegt von Christoph Paul Freier
aus
La Garenne-Colombes, Frankreich
Betreuer: Prof. Dr. rer. nat. Udo Jeschke
Zweitgutachter: Prof. Dr. rer. nat. Vigo Heissmeyer
Drittgutachter: Prof. Dr. rer. nat. Roland Kappler
Viertgutachter: PD Dr. rer. nat. Reinhard Obst
Dekan: Prof. Dr. med. dent. Reinhard Hickel
Il faut imaginer Sisyphe heureux.
theses.
Nonetheless, this document was handmade using theXeTeX typesetting sys-tem created by the Non-Roman Script Initiative and the memoir class created by Peter Wilson.
Analyses were designed and performed using Rand occasionally powered by
Python. The body text is set 12pt with Adobe Caslon Pro. Other fonts include
Optima RegularandEnvy Code R.
CPF
Printed in Münster • July 4, 2016 Contact me:christoph@freier.fr
Preface
T
his thesis embraces all the efforts I’ve made during the last four years as a Ph.D.student at the Ludwig-Maximilians-Universität of Munich. Or, actually, even since I eventually decided to study pharmacy, ten years ago. It has been quite a long journey since these warm days of July 2005.I can remember this very first day of July were we went, my mother and I, for the first time to the faculty of pharmacy. I thought it would be to late or we couldn’t make it. Indeed, alone, I certainly wouldn’t have even reached the secretariat.
I can remember the first days in a pharmacy, luckily direct nearby Thomas’ one. Also, every cup of tea we shared, me and Bogdann, and the innumerable and endless nights at Morgan’s place. Guys, you made everything possible.
I can remember the first experience of what laboratory work means. I learned there everything I needed to successfully conduct this work and it has been the very beginning of my love for research. Without Marc, this thesis wouldn’t exist today.
I can remember the first stays in Germany and the relaxing sailing trips with Jan. The laborious learning of German with Stefanie, Marie, and Eva. Looking back, it’s easy to realize how much I learned and how important it was.
I can remember this summer evening at the Königin 43 with my brother Niels. Only armed with our laptops and two Weißbiere, he taught me how to code and to run my very firstRanalyze. It certainly has been one of the most productive evening of this work.
I can remember that I also have two other brothers, Carl and Victor, both simi-larly important. I’m sorry I missed so much in France, while they grew up, became adults, studied, worked, lived and experienced so much. I’m so proud of them, too. What I can’t remember is my grandfather saying in front of me, “Et dire qu’un jour, ça passera son bac.” I was only a new born at that time. Now that I’ve grew up, his prediction became true, even outperformed. I’m sure he would have been proud, if he could have read this. This work is dedicated to his loving memory.
Christoph Paul Freier München July 12, 2015
Abstract
Purpose: To assess the clinical significance of different regulatory T cells (Treg) attracting chemokine intratumoral expressions in breast and ovarian cancer patients with available long-term follow-up (15 years). We investigated the prespecified hypothesis that the expression of Treg specific chemokines in cancer tissue predicts theoverall survival (OS).
Patients and Methods: We screened all so far known chemokines of the CC-family for their capacity to selectively attract Treg in vitro. Three chemokines (CCL1, CCL22, and CCL27) with selective action on Treg migration were stained using extensively characterized antibodies. The numbers of positive cells in tissue microarray cores from ovarian cancer and invasive breast cancer were computer-aided determined.
Results: Within all analyzed chemokines, only CCL1, CCL22 and CCL27 selectively attracted Treg in vitro. All three chemokines were strongly ex-pressed in most ovarian and breast cancer tissues. Moreover, there was a significant relationship between Treg infiltration in tumors and CCL1- and CCL27-expressing cell total numbers, whereas no association was seen for CCL22. High numbers of CCL1- or CCL27-positive cells identified patients with shorter OS.
Conclusion: Our findings indicate that quantification of intratumoral Treg attracting chemokines in ovarian and breast cancer is valuable for assessing disease prognosis. Unlike conventional clinicopathologic factors, high expres-sion of certain chemokines can identify patients at risk of death over 15 years. CCL1 and CCL27 represent novel markers for identifying effect of immune response and tumor escape as well as patients who may benefit from immuno-therapy. Such chemokines may gain to be considered together and could rep-resent important therapeutic targets.
Key words: regulatory T cells, chemokine, CCL1, CCL27, ovarian cancer,
breast cancer, clinical outcome, biomarker, therapeutic target, cancer immuno-therapy.
Zusammenfassung
Zielsetzung: Untersuchung der klinischen Signifikanz von intratumoral ex-primierten Chemokinen bei Mamma- und Ovarialkarzinompatientinnen mit Langzeit-Follow-up (15 Jahre) und deren Einfluss auf die Infiltration regula-torischer T-Zellen (Treg). Untersucht wurde die Hypothese, dass die Expres-sion Treg spezifischer Chemokine in Karzinomgeweben das Gesamtüberleben beeinflusst.
Patientenkollektiv und Methode: Überprüft wurden alle bisher bekannten Chemokine der CC-Familie auf die selektive Kapazität Treg Migration in
vitro auszulösen. Drei Chemokine (CCL1, CCL22 und CCL27), die die Treg
spezifisch anlocken, wurden mit getesteten Antikörpern gefärbt. Die Anzahl positiver Zellen in Microarray-Gewebsschnitten von Ovarial- und invasiven Mammakarzinomen wurden mit Hilfe eines Computerprogramms bestimmt. Ergebnisse: Von allen untersuchten Chemokinen, lockten nur CCL1, CCL22 und CCL27 Treg in vitro an. Die meisten Ovarial- und Mammakarzinom-gewebe zeigten eine starke Expression dieser drei Chemokine. Weiterhin bestand ein signifikanter Zusammenhang zwischen Treg-Infiltration ins Tu-morgewebe und der Gesamtanzahl CCL1- sowohl als auch CCL27- exprim-ierender Zellen, während kein Zusammenhang mit CCL22 gefunden wurde. Eine hohe Anzahl an CCL1 oder CCL27 positiver Zellen ging einher mit verkürztem Gesamt-überleben.
Schlussfolgerung: Unsere Ergebnisse weisen darauf hin, dass Treg anlock-ende Chemokine in Ovarial- und Mammakarzinomen von prognostischer Be-deutung sind. Im Gegensatz zu konventionellen klinikopathologischen Fak-toren, identifiziert eine hohe Expression bestimmter Chemokine Patientinnen mit erhöhtem Risiko innerhalb der nächsten 15 Jahre an dem Karzinom zu ver-sterben. CCL1 und CCL27 stellen neue Marker zur Untersuchung der Immu-nantwort und der Reaktion des Tumors diese zu umgehen dar. Des Weiteren dienen diese beiden Chemokine zur Identifizierung der Patientinnen, die von einer Immuntherapie profitieren würden. Eine Kombination von CCL1 und CCL27 könnte in Zukunft ein wichtiges therapeutisches Target darstellen.
Contents vii
1 Introduction 1
1.1 The immunology of tumors . . . 1
1.1.1 A short story of the immune regulation . . . 1
1.1.2 Role of the immune system in cancer . . . 3
1.1.3 Clinical impact of the immune contexture . . . 3
1.1.4 Emergence of the immunotherapy . . . 4
1.2 Regulatory T cells . . . 5
1.2.1 About their role in the immunoediting process . . . 5
1.2.2 Characteristics of regulatory T cells . . . 6
1.2.3 Regulatory T cell markers . . . 6
1.2.4 Regulatory T cell mechanisms of action . . . 7
1.3 Chemokines and their receptors . . . 8
1.3.1 Discovery . . . 8
1.3.2 Nomenclature . . . 9
1.3.3 Structure and classification of chemokine . . . 10
1.3.4 Classification of chemokine receptors . . . 13
1.4 Goals of this work . . . 15
1.4.1 Epidemiology of gynecological tumors . . . 15
1.4.2 Current problematic . . . 15
1.4.3 Working hypothesis . . . 16
1.4.4 Purpose of this work . . . 16
2 Materials and Methods 17 2.1 Lymphocyte acquisition and manipulation . . . 17
2.1.1 Peripheral blood mononuclear cell isolation . . . 17
2.1.2 Lymphocyte maintenance and culture . . . 17
2.1.3 Migration assay . . . 18
2.1.4 Flow cytometry analysis . . . 18
2.2 Chemokine receptor mRNA level expressions . . . 18
2.2.1 Cell sorting . . . 18
2.2.2 mRNA isolation . . . 19
2.2.4 RT-qPCR analysis . . . 20
2.3 Histopathology . . . 21
2.3.1 Breast cancer patient recruitment . . . 21
2.3.2 Ovarian cancer patient recruitment . . . 22
2.3.3 Immunohistochemistry . . . 22 2.3.4 Quantification of immunohistochemistry . . . 23 2.4 Statistical analysis . . . 23 2.4.1 Statistical program . . . 23 2.4.2 Statistical tests . . . 25 2.4.3 Survival analysis . . . 25
2.4.4 Kaplan-Meier curves with ggplot2 . . . 25
3 Results 27 3.1 Chemokine screening . . . 27
3.1.1 Screening among the CC-chemokine family . . . 27
3.1.2 CCL1, CCL22 and CCL27 specifically attract regulatory T cells . . . 29
3.1.3 Internal control and validation of our targets . . . 29
3.2 Expression of chemokines in ovarian cancer . . . 31
3.2.1 First description of CCL1 expression in ovarian cancer . . 31
3.2.2 CCL1 significantly more expressed in cancer tissues . . . . 33
3.2.3 Association of CCL1 expression with clinicopathologic pa-rameters . . . 34
3.3 Expression of chemokines in breast cancer . . . 35
3.3.1 CCL1, CCL22 and CCL27 all expressed in breast cancer . 35 3.3.2 CCL1 and CCL27 significantly more expressed in cancer tissues . . . 36
3.3.3 FoxP3+ cells significantly more infiltrated in cancer tissues . 36 3.3.4 Association of CCL1, CCL22 and CCL27 expressions with clinicopathologic parameters . . . 37
3.4 Chemokines correlate with worse overall survival . . . 39
3.4.1 CCL1 expression predict poor patient outcome in ovarian cancer . . . 39
3.4.2 Prognostic significance of CCL1-expression in ovarian cancer 41 3.4.3 CCL1 and CCL27 expression predict poor patient outcome in breast cancer . . . 43
3.4.4 Prognostic significance of CCL1- and CCL27-expression in breast cancer . . . 45
4 Discussion 47 4.1 Chemokines specifically attracting regulatory T cells . . . 47
4.1.1 A brief history of chemokines in cancer . . . 47
4.1.2 Our candidates: CCL1, CCL22 and CCL27 . . . 48
4.2 The case of ovarian cancer . . . 49 4.2.1 CCL1 produced by tumor and immune cells . . . 49 4.2.2 Both expressions correlated with reduced overall survival . 49 4.2.3 CCL1 could serve as a new biomarker in ovarian cancer . . 49 4.3 The case of breast cancer . . . 50 4.3.1 CCL22 similarly produced in tumor tissue . . . 50 4.3.2 CCL1 and CL27 produced in tumor tissue . . . 50 4.3.3 CCL1 and CCL27 could be targeted in immunotherapy . 51 4.4 Conclusion . . . 51
4.4.1 Comparison between ovarian and breast cancer chemokine expressions . . . 51 4.4.2 Limitation of our investigations . . . 52 4.4.3 Concluding remark . . . 52 5 Appendix 55 5.1 Curiculum vitæ . . . . 55 5.2 Eidesstattliche Versicherung . . . . 59 List of Figures 61 List of Tables 63 List of Acronyms 65 Bibliography 69 Index 79
—It is by no means inconceivable that small accumulations of tumour cells may develop and, because of their possession of new antigenic po-tentialities, provoke an effective immunological reaction with regression of the tumour and no clinical hint of its existence.
Macfarlane Burnet, immunologist, 1957
1.1 The immunology of tumors
1.1.1 A short story of the immune regulation
T
he balance between effector response and mechanisms of immune regulation isone of the fundamental features of the immune system: the ability to correctly respond to any external aggression or infectious microorganism avoiding destruction of self-tissues determines its role [1]. This original paradigm was claimed more than fifty years ago by Medawar et al. and derived nowadays to include the pharmacologi-cal induced state of tolerance against any deliberately introduced antigen, as it is the case for a transplanted organ. From an evolutionary point of view, the critical im-portance of the immune tolerance is illustrated by the multitude of non-redundant mechanisms [2].It is currently well accepted that immunological tolerance is mediated by two categories of mechanism: central and peripheral. After a first cellular selection in the thymus, the functionality and efficiency of the immune system is highly de-pendent on complex series of cellular and cytokine mediated interactions between effector and regulatory cells. In the periphery, effector but also regulatory cells have to be recruited both at the right place and at the right time, via complex chemokine networks coordinating cell traffics.
Figure 1.1 – Tumor anatomy showing the immune contexture of cancer with features of the
immune contexture and the distribution of different immune cells according to Fridman et al.. DC, dendritic cell; MDSC, myeloid-derived suppressor cell; Treg, regulatory T cell; NK cell, natural killer cell; FDC, follicular dendritic cell; TLS, tertiary lymphoid structure.
These chemokines consist of small chemotactic cytokines characterized by criti-cally positioned cysteine residues that are recruiting distinct cell types according to their corresponding receptors [3, 4]. Corresponding chemokine receptors are spe-cifically regulated in regard of cell type, stage of activation and differentiation [5]. Given the central role of chemokines in the regulation of cell migrations, it was obvious that they also play a pivotal role in the regulation of the peripheral immune response. They allow the regulation of the adaptive immune response by recruiting effector cells, but also immunoregulatory cells. So far, over 50 chemokines and 22 chemokine receptors have been identified and their central function in the coordi-nation of the immune system is now firmly established [6].
1.1.2 Role of the immune system in cancer
Many of the defenses against the appearance of cancerous growths are inherent in cells. The most obvious one lying in the different controls imposed by the apop-totic machinery. During the past decades, however, several authors linked human cancer progression with immune cell infiltration [7, 8, 9]. This followed the first observations realized in mice by Hanahan and Weinberg, which had led them to add “avoid immune destruction” as a hallmark of cancer [10]. In fact, tumor en-vironments are an intricate network of epithelial cells, vascular and lymphatic ves-sels, cytokines, chemokines and infiltrating immune cells. All immune cells can be found in a tumor, including dendritic cells (DC), macrophages, natural killer cells, naive and memory lymphocytes, B cells and various subsets of T cells – Figure 1.1, p. 2 [11]. Moreover, these different cell types seem to play a role which can vary according to the tumor nature and his localization: sometimes favorable, or in some other cases, paradoxically accelerating the tumor growth and reducing the OS of patients [12, 13].
The demonstration of the infiltrated cell influence on the clinical outcome of cancer conducted to the creation of the field of tumor immunology and already led to novel opportunities of prognostic marker identification and therapeutic options. Histopathological studies of human tumors have shown that variable numbers of infiltrating immune cells are found in different tumors of the same type, and are found at different localizations within and around a tumor; this distribution also varies between tumor types [14]. This is underlying the possibility of different roles for these immune cell populations, in respect to their localizations in tumors, but also the complexity of the role played by the immune system in cancer.
The observation of variable densities and localizations of these immune cells between tumors of the same cancer type in different individuals tends to prove the orchestrated fine migration of these cells. Considering the role of chemokines in cell migration, it was not surprising to find overexpressed chemokine synthesis in several tumor types, possibly responsible for the accumulation of immune cells [15, 16]. These issues are far from being completely answered, but analyses of the cytokine and chemokine milieu associated with a tumor immune contexture are accumulating. Moreover, this suggests that the immune system could indeed represent effective defenses that prevent the appearance of tumors.
1.1.3 Clinical impact of the immune contexture
One century ago Paul Ehrlich proposed that the immune system is programmed to avoid the generation of autoreactive immune responses and termed this aversion to autoreactivity “horror autotoxicus”. Ehrlich’s observations that goats could make anti-bodies against the blood components of other goats, but not against their own blood, represented the first evidence of immunological self-tolerance. However, nowadays, it clearly appears that an absence of immune response is as much deleterious for the organism.
Correlations between tumor levels of immune cell infiltration and clinical out-come have been investigated in many cancers and results are still a matter of debate since they present some contradictions. A strong lymphocytic infiltration has been reported to be associated with good clinical outcome in many different tumor types, including melanoma, head and neck, breast, bladder, urothelial, ovarian, colorectal, renal, prostatic and lung cancer [11]. Precisely, high densities of cluster of differen-tiation (CD)3+ T cells, CD8+cytotoxic T cells and CD45RO+memory T cells were commonly associated with better prognosis [17, 18].
In contrast, analysis of the CD4+ T cell population infiltration often shows an opposite effect on OS. Most conflicting are the different data forT helper cells (Th)2, Th17 and Treg, linked with difficult interpretations [11]. This is presumably due to the creation of an immunosuppressive micro-environment, inhibiting any chance of a correct immune response against tumor cells and thus leading to accelerated tumor growth and reduced OS. This immune contexture in tumor could serve to predict therapeutic responses, as shown in the colorectal cancer and the breast cancer. Considering the number of infiltrated cytotoxic CD8+ cells and memory CD45RO+ cells, Denkert et al. and Tosolini et al. could finely predict the risk of relapse when no other actual markers could do it [18, 19]. This of course, could be of paramount importance in patient clinical management, but also allows to target the accumulation of unfavorable immune cells; thus opening the way to interesting new therapeutic opportunities of treatment.
1.1.4 Emergence of the immunotherapy
During the last two decades, cancer treatments evolved from relatively nonspe-cific cytotoxic agents to selective, mechanism-based therapeutics. Although cancer chemotherapies initially were compounds that killed rapidly but unspecifically divid-ing cells, improved understanddivid-ing of cancer pathogenesis has given rise to new treat-ment options, including targeted agents and cancer immunotherapy. Among these modern targeted approaches, immunotherapy aims to stimulate a host immune re-sponse that effectuates long-lived tumour destruction to improves clinical outcome, reduces toxicities and frequently acquired resistances. Significant advances in target-ing the immune response were made and have led to cell-based vaccine composed of autologousantigen presenting cells (APC)that have been exposed to a recombinant protein consisting of granulocyte-macrophage colony-stimulating factor (GM-CSF) fused to a protein expressed by cancer cells. Upon administration, the vaccine may stimulate an antitumor T-cell response against tumor cells expressing this protein.
Secondly, tumour-specific T cells must differentiate into effector T cells, which requires a combination of signals from both theT cell receptor (TCR)and several co-stimulatory molecules [20]. Co-co-stimulatory signals are delivered through multiple transmembrane proteins of the B7 andtumour necrosis factor receptor (TNFR) fam-ilies, as well as receptors for some cytokines, such as interleukin (IL)-12. Agonistic antibodies to these molecules can enhance co-stimulation to augment anti-tumour immunity [21, 22]. However, T cells must avoid negative regulatory signals (known
as immune checkpoints) that dampen their activation or that induce tolerance pro-grammes such as anergy or exhaustion [23, 24]. Blockage of cytotoxic T lymphocyte-associated antigen 4 (CTLA4)andprogrammed cell death protein 1 (PD1), both major negative co-stimulatory molecules that are expressed on activated T cells, are now possible and has demonstrated encouraging anti-tumour effects in initial clinical testing [25, 26].
A major impediment to the therapeutic efficacy of anti-tumour T cells is the immunosuppressive tumour micro-environment [27]. Diverse mechanisms operate, including the production of inhibitory cytokines such as IL-10 and tumor growth fac-tor (TGF)-β, the expression of negative co-stimulatory ligands such asprogrammed cell death ligand 1 (PDL1), and the presence of regulatory lymphocyte and myeloid cell populations. The identification of agents that attenuate these suppressive net-works might substantially increase the efficacy of immunotherapies [28].
1.2 Regulatory T cells
1.2.1 About their role in the immunoediting process
Tumors are often infiltrated by various inflammatory immune cells creating a milieu that may function to either stimulate or inhibit cancer growth [29]. The chronic inflammation response found in many cancers creates an environment rich in sub-stances that promote angiogenesis and cell proliferation, very similar to wound heal-ing processes [30]. Evidence from murine studies strongly suggest, that adaptive immune cells are attracted and can directly eliminate tumor cells [31, 32].
Immunosurveillance, however, can fail to completely kill nascent tumors – due to poor immunogenicity of tumor cells or immune escape mechanisms – and there-fore favorises the immunoediting process which eventually ends in tumor escape [8]. The analysis of the location, density, functional orientation of different immune cell populations, and their respective clinical impacts have been investigated in many cancers [11]. The ultimate goal is to allow the identification of components of the immune contexture that are beneficial, as well as those that are deleterious, to pa-tients.
A strong lymphocytic infiltration, particularly CD8+ cytotoxic T lymphocytes, has been reported to be associated with responses to neoadjuvant chemotherapy and good clinical outcome in breast cancer [18, 33]. CD4+ CD25+ CD127- Treg play a pivotal role in the control of immune responses [34, 35]. Treg normally function to protect tissues from autoimmune diseases by suppressing self-reactive cells, including CD8+ cytotoxic T lymphocytes, B cells and natural killer cells [36, 37, 38]. Additionally, they have an important role in the immunoediting process, enabling the tumor to elude the host antitumor immune response, because of their inhibiting action on cytotoxic activity of CD8+and natural killer cells [36, 37].
1.2.2 Characteristics of regulatory T cells
Treg represent a subpopulation of CD4+ lymphocytes that suppress the immune response at different levels. Human peripheral blood CD4+ CD25+ forkhead box P3 (FoxP3)+T cells, or Treg represent 4-8% of human CD4+T cells [39]. They have been characterized as suppressor T cells after the observation by Sakaguchi et al. of induced organ-specific autoimmune disease in mice depleted for this particular population [40].
The expression of CD25 (also known as IL-2 receptor-α) appears during the transition of the CD4+ CD8+ T cell to the CD4+ CD8- T cell, therefore taking part in normal thymocyte differentiation [41]. This reflects the activation of cells during a process termed “altered negative selection”: future Treg probably express a TCR with an intermediate affinity for self, to low for negative selection, but to high to allow them to pass through to the periphery [42].
Minor differences are observed in respect to Treg population of “naturally occur-ring”, which are the thymus derived cells naturally isolated from peripheral blood and “induced”, generated in the lymph node. However, it is widely accepted that they are both hyporesponsive and suppressive, as a result of the activation process, acting via an APC-dependent mechanism [43]. Thus in normal process, Treg mod-ulate immune responses through selective migration and accumulation at different sites where regulation is required. Accumulative amount of data has demonstrated the role in controlling Treg migration of distinct signals from chemokines/chemokine receptors and integrins/integrin ligands in recruiting Treg in human pathologies and for the homing of these cells, indispensable for them to acquire their full activation state [44, 45]. In the mouse, direct evidences of roles for migration and functionality played by chemokines and integrins was also shown in vitro and in vivo [46, 47].
1.2.3 Regulatory T cell markers
Despite the immunologists’ obsession to classify all immune cell types, the identifi-cation of discriminatory cell-surface markers for the characterization and isolation of Treg remains problematic. This could account for the sometime confusing re-sults found in several Treg studies relative to their implication in cancer and is even more surprising considering that we already dispose of a large panel of Treg mouse markers. If traditionally, mouse and human Treg have been characterized as CD4+ CD25+, the purity of isolated human Treg remains an issue mainly because other human T cells also upregulate CD25 when activated [48].
One significant advance in the study of mouse and human Treg has been the identification of FoxP3 as key regulator of Treg development and function [49, 50]. Elegant mouse and human genetic studies demonstrated that mutation in the FoxP3 gene were linked to the autoimmune manifestations observed in the scurfy mouse that carry a spontaneous loss-of-function mutation or a deletion of FoxP3 and human immune dysregulations like polyendocrinopathy, enteropathy and X-linked syndrome disease. But again, if it appears useful in mice, many activated
Figure 1.2 – Regulatory T cell mechanisms of action according to Vignali et al.. Four basics mode
of action are here illustrated (a) via inhibitory cytokines, (b) via the cytolysis of targeted cells, (c) via metabolic disruption and (d) via the inhibition of dendritic cells. TGF-β, transforming growth factor-β; IL, interleukin, A2R, adenosine receptor; MHC, major histocompatibility complex.
non-regulatory human T cells transiently up-regulate FoxP3 under activation [48]. Moreover, FoxP3 is intracellular and could not serve routinely to Treg isolation so that the search for specific surface markers has continued in earnest with the discov-ery of the downregulation of CD127 (also known as IL-7 receptor) on Treg and the 90% correlation of the three membrane markers (i. e., CD4+, CD25+and CD127-) and FoxP3 expression [35].
Finally, it is possible that Treg, like other T cell populations, are composed of heterogeneous subtypes, like suggested by Sakaguchi et al. [51]. They divided Treg into three populations based on the expression of FoxP3 and CD45RO and showed that these populations suppressed other cells via different mechanisms and not all with the same intensity. Thus it appears probable that the search for better mem-brane markers is not yet finished.
1.2.4 Regulatory T cell mechanisms of action
Defining how do these cells work is crucial to provide insight into the control pro-cesses of peripheral tolerance and could also indicate potentially important thera-peutic targets. Treg require TCR triggering to become fully functional, but once activated they suppress T cells blindly, with no regard to the antigen specificity of targeted cells [43].
of action: (a) suppression by inhibitory cytokines, (b) suppression by cytolysis, (c) suppression by metabolic disruption and (d) suppression by modulation of DC mat-uration and/or function – Figure 1.2, p. 7 [52]. To date, it is difficult to say if Treg functions are mediated by a single overriding suppressive mechanism or by multi-ple, redundant or not, mechanisms. It is, however, more likely that Treg suppressive functionality comes from multiple one, since none of the currently known mecha-nisms seems to be exclusive. Again, the choice for suppression mechanism(s) to use – dependent of the background, the context or the type of target cells – remains unclear, underlying the need of more research in this domain.
Several studies already suggested that Treg infiltrated in tumors may adversely affect prognosis via their immunosuppressive action. Increased tumoral Treg num-ber have been correlated with poor OS in breast cancer, melanoma, lung carcinoma, and hepatocellular carcinoma [14, 53, 54, 55]. Importantly, the high number of Treg present in tumor environment raises the question of their recruitment.
1.3 Chemokines and their receptors
1.3.1 Discovery
The first description of directed, transendothelial leucocyte migration and accumu-lation into organs in case of inflammation was made more than one century ago by August Waller (1846) and Julius Cohnheim (1867). The first hypothesis, that leucocytes could phagocytose bacteria and thus should have to migrate to inflamed tissues, eventually in response to chemical agents, was postulated a couple of years later by Ilja Metchnikoff (1891). But it took almost one century to have the first in vitro demonstration of this phenomenon, when in the 60’s and 70’s the migra-tion was observed via migramigra-tion assays using the first chamber chemotaxis transwell plates.
The next step was done in the 80’s, when several teams discovered and char-acterized a new chemoattractant polypeptide named monocyte-derived neutrophil chemotactic factor (MDNCF) or neutrophil-activating peptide (NAP). One year af-ter, this chemokine was cloned, sequenced and renamed IL-8, first chemotactic cytokine [56]. With help from sequence homologies and hybridization methods other members of this new family were quickly identified. A few years later, on the “3rdInternational Symposium on Chemotactic Cytokines” it was decided to rename this cytokine subfamily after their chemotactic-cytokine effects into chemokine.
The list of chemoattractant quickly grew from C5a, through platelet activating factor (PAF)to the more than 50 chemokines or chemoattractants known today [3]. All these different chemoattractants were observed to bind to specific surface mem-brane receptors, all members of theG protein-coupled receptors (GPCR)family. In the mean time, a lot of other signal polypeptides were discovered, with the result that it soon began urgent to order all these new cytokines in subfamily and to give
them an international nomenclature. Originally most of them were named with different names, according to their different effects on cells.
1.3.2 Nomenclature
In 2000, a system of nomenclature was introduced in which each ligand and receptor were respectively identified by its subfamily and given an identifying number. For example, CCR4 refers to the chemokine receptor of the CC-family number 4, and binds to CCL17 and CCL22 (chemokine ligand of the CC-family, number 17 and 22 respectively). Redundancy between different ligands and the same receptor or on the contrary, between different receptors and the same ligand is considerable. If very little is known about this lack of specificity, it may be an explanation for the quite normal phenotypes of knockout (KO)mice for most inflammatory chemokines or chemokine receptors unless challenged with inflammatory conditions or faced with pathogens [57]. On the other hand, this redundancy could allow fine tuning of immune responses, function of combination of ligands, receptors, but also other molecules, possible oligomerization, glycosaminoglycans, etc.
CCL#, CXCL#, CL# and CX3CL# refer to the four families of chemokine ligands. C stands for a cysteine and X for a non-cysteine amino acid, L for ligand, and # denotes the identifying number. The larger chemokine family, composed by the CCLs or β-chemokines, is characterized by two adjacent cysteines, near to the N-terminal end of the protein – Table 1.1, p. 10. 27 chemokines have been reported in this group, named CCL1 to CCL28 (CCL9 and CLL10 being the same one). Notably a small number of them possess six cysteines: CCL1, CCL15, CCL21, CCL23, CCL28.
In the second largest family, composed by the CXCLs or α-chemokines, the two first cysteines are separated by one amino acid, represented in the name with an “X” – Table 1.2, p. 11. 17 of them have been described, subdivided in two categories: those with a specific amino acid sequence ELR (glutamic acid-leucine-arginine) immediately before the first cysteine of the CXC motif, and those without.
XCL1 and XCL2 are the two chemokines characterized, in the third family, the XCLs or γ-chemokines, and are characterized by only two cysteines: one N-terminal cysteine and one cysteine downstream – Table 1.2, p. 11. These chemo-kines are presumed to attract T cell precursors into the thymus [5].
Finally, CX3CL1 is the only chemokine discovered in the fourth family, CX3CL or δ-chemokine – Table 1.2, p. 11. In the same fashion as for the CXC chemokines, the two first cysteines are separated, in this case indeed, by three amino acids. It is both secreted and tethered to the surface of the cell that expresses it, thereby serving both as chemoattractant and as an adhesion molecule [3].
It is of importance to mention, that not all of these chemokines are present both in mouse and human, underlying the difference between the two systems, but also the functional redundancy of chemokines.
Name Old name Produced Binds on Receptor(s) CC (β)
CCL1 i I-309 M, T M, act.Th2 CCR8
CCL2 i MCP-1 M, Fib, En M, T, Ba, Fib CCR2,3
CCL3 i MIP-1α M, T, NK, N, DC, Fib, En M, T, DC CCR1,5
CCL4 i MIP-1β M, T, NK, N, DC, Fib, En M, DC CCR5
CCL5 i RANTES M, T, Fib, Epi, En, P, Eos M, T, DC, Eo CCR1,3,5
CCL7 i MCP-3 M, Fib, P M, T, NK, DC CCR1,2,3,5
CCL8 i MCP-2 N, Fib M, T, NK, Eo CCR2,3
CCL11 i Eotaxin-1 M, Eos, Fib, resp. Epi Eo, Ba, Th2 CCR3,5
CCL13 i MCP-4 M, L, En, Epi / Lung, Gut M, T, DC, Eo CCR2,3,5
CCL14 h HCC-1 S, G, Li, Mu M, Eo CCR1
CCL15 h HCC-2 L, NK, M, DC / Liver, Gut T, M, Eo CCR1,3
CCL16 h LEC M / Liver M, L CCR1
CCL17 m TARC T, DC / Thymus, Lung, Gut Thy, Th2 CCR4
CCL18 h PARC M, DC / Lung, LN, Thymus T CCR8
CCL19 h ELC DC / Thymus, LN T, B, DC, Thy CCR7
CCL20 m LARC Epi, M, DC, T / LN, Lung T,DC CCR6
CCL21 m SLC En / Thymus, Spleen, LN T, B, DC, Thy CCR7
CCL22 m MDC M, DC / Thymus, LN, Gut Th2, NK, DC CCR4
CCL23 i MPIF-1 DC / Pancreas, Muscle M, L CCR1
CCL24 i Eotaxin-2 M / Lung, Skin Eo, Ba, Th2 CCR3
CCL25 h TECK Epi, DC / Thymus, Liver, Gut Thy, T, M, DC CCR9
CCL26 i Eotaxin-3 Fib, En / Skin, Heart, Ovar Eo, Ba CCR3
CCL27 h CTACK Epi / Skin, Placenta T, Fib, En CCR10
CCL28 i MEC Epi / Brust, Colon T, Eo CCR3,10
i, pro-inflammatory; h, homeostatic; m, mixed function
Table 1.1 – Classification of human CC-chemokines. MCP-cluster in blue, MIP-cluster in red. B, lymphocyte B; Ba, basophil; DC, dendritic cell; Eo, eosinophil; En, endothelium; Epi, epithelium; Fib, fibroblast; Hep, hepatocyte; L, lymphocyte; M, monocyte/macrophage; N, neutrophil; NK, natural killer cell; P, platelet; T, lymphocyte T; Th1/2, type 1/2 helper T cell; Thy, thymocyte; LN, lymph node
1.3.3 Structure and classification of chemokine
Chemokines are small ligands, 70-130 amino acids and 8-12 kDalton (Da)proteins (with the notable exception of CXCL16 and CX3CL1) that contain 1-3 (mostly 2) disulfides, with critical roles in cell migration during immune surveillance, in-flammation and development. They interact with GPCR on target cells and cause conformational changes that trigger intracellular signaling pathways involved in cell movement and activation and are classified as chemokine based on shared structural characteristics, small size and the presence of four cysteine residues (except for the γ family and a few exceptions in the β family) in conserved locations that inter-act with each other in pairs to create a Greek key shape that is a charinter-acteristic of chemokines.
Although the sequence homology is highly variable, ranging from less than 20% to over 90%, the tertiary structure is remarkably conserved – Figure 1.3, p. 12. This tertiary structure always presents a disordered N-terminal of 6-10 amino acids func-tioning as key signaling domain. The N-terminal has a critical role in receptor activa-tion, and N-terminus truncations can render chemokines inactive or even able to act
Name Old name Produced Binds on Receptor(s) CXC (α) CXCL1 i GRO-α M, Epi, En N, L, M, En CXCR2 CXCL2 i GRO-β M, Epi, En N, L, M, En CXCR2 CXCL3 i GRO-γ M, Epi, En N, L, M, En CXCR2 CXCL4 h PF-4 P Fib, En CXCR3 CXCL5 i ENA-78 M, N, Epi N CXCR1,2
CXCL6 i GCP-2 M, Fib / Heart, Lung, Kidney N CXCR2
CXCL7 i NAP-2 P, M, N N CXCR2
CXCL8 i IL-8 M, L, N, Fib, Epi N, T CXCR1,2 CXCL9 i Mig M, Fib, Hep, IFN / Spleen, Liver T, NK CXCR3 CXCL10 i IP-10 M, IFN / LN, Thy T, NK CXCR3 CXCL11 i I-TAC M, IFN / Spleen, Pancreas T CXCR3,7
CXCL12 h SDF-1 Fib / Spleen, Gut T, B, DC CXCR4,7
CXCL13 h BCA-1 DC, En / Spleen, LN, PeyerPatch B CXCR5
CXCL14 i BRAK all M, DC ?
CXCL16 i SR-PSOX DC / LN, Spleen, Lung T, NK, DC CXCR6
CXCL17 i DMC Stomach, Trachea D, M ?
C (γ)
XCL1 m Lympotactin T / Thymus, Spleen T, NK XCR1
XCL2 m MCP-1 T L XCR1
CX3C (δ)
CX3CL1 i Fractalkine En / Brain, Lung, Heart M, T, NK CX3CR1
i, pro-inflammatory; h, homeostatic; m, mixed function
Table 1.2 – Classification of all other human chemokines. GRO-cluster in green and IP-10-cluster in yellow. B, lymphocyte B; DC, dendritic cell; En, endothelium; Epi, epithelium; Fib, fibroblast; Hep, hepatocyte; IFN, IFN-γ stimulated cell; L, lymphocyte; M, monocyte/macrophage; N, neu-trophil; NK, natural killer cell; P, platelet; T, lymphocyte T; Thy, thymocyte; LN, lymph node
as antagonists. Most of them are secreted from the cell, with two notable exceptions: CX3CL1 and CXCL16, which also can stay on the extracellular cell membrane [3]. To produce directional migration signals for cell migration, concentration gradients of chemokines must be formed, since cells exposed to a uniform concentration of a chemokine move in random patterns rather than directionally. In vivo, the establish-ment of a gradient is thought to involve binding to glycosaminoglycans; conditions that could be mimiced in vitro by the chamber chemotaxis migration assay.
Chemokines can also be classified into three groups, according to their physi-ological roles. Inflammatory chemokines are inducible in organs in processes like immune response or inflammation (inflammation mediated by cytokines, bacterial toxins, etc). They attract specialized leukocyte populations or subpopulations (e.g., neutrophil, monocyte, basophil, eosinophil, effector T cell, DC) to the inflamed lo-calization and allow the immune system to respond various aggressions. They also regulate the expression of genes in those cells, acting as transcription factor or via stabilization of messenger RNA (mRNA). The multitude of inflammatory chemo-kines allows fine regulations of the immune response. This redundancy could be explained, from an evolutionary point of view, by multiple gene duplications. There are to date four clusters: MCP-cluster and MIP-cluster of the chromosome 17q11-17q12, GRO-cluster of the chromosome 4q13 and IP-10-cluster of the
chromo-Figure 1.3 – Schematic representation of the chemokine key functional regions. All chemokines
have a similar tertiary structure comprising a disordered N-terminal of 6-10 amino acids followed by a long loop (known as the N loop), a 310helix, a three-stranded beta-sheet, and finally a
C-terminal alpha helix.
some 4q21. These chemokines present highly conserved tertiary structures and act mostly on the same receptor with a similar action. Homeostatic chemokines are constituently expressed in the primary and secondary lymphoid organs and control the migration and localization of T and B cells, as well as DC. This expression is particularly important for the maturation of these cells and for the adaptive immune response. Immune surveillance also involves migration of lymphocytes in response to homeostatic chemokines. Certain chemokines are continuously expressed in tis-sues such as the skin, intestinal mucosa, and lungs to promote constant monitoring of these at-risk areas. Lastly, some chemokines are included in a third group, re-flecting their mixed function, depending on the biological context or pathological state. Expression of inducible chemokines is often triggered by inflammatory me-diators such astumor necrosis factor (TNF)-α,interferon (IFN)-γ, microbial products, or trauma. These chemokines have roles in both innate and adaptive immunity in
Figure 1.4 – Schematic representation of the chemokine receptor key functional regions.
Muta-genesis and chimeric receptor studies have identified regions of chemokine receptors important for ligand binding, receptor activation, and internalization, although specific sequences involved in signaling differ between different chemokine receptors.
response to infections, tissue damages, and other physiological abnormalities. Their expression is temporary, until resolution of the situation.
1.3.4 Classification of chemokine receptors
The chemokine receptors are all class A GPCR, presenting three intracellular and three extracellular hydrophilic loops – Figure 1.4, p. 13. They are commonly 340-370 amino acids long and about 40 kDa proteins. The N-terminal domain is ex-posed outside the cell and binds to chemokine(s). The C-terminal domain, con-taining serine and threonine residues that act as phosphorylation sites, is coupled with the G protein. Conserved extracellular cysteine disulphide bridges stabilize the tertiary structure.
Name Expressed on Ligand(s) CXC receptor CXCR1 N CXCL6, 8 CXCR2 alt N, En CXCL1, 2, 3, 5, 6, 7, 8 CXCR3 alt Th1 CXCL9, 10, 11 CXCR4 All CXCL12 CXCR5 B, T CXCL13 CXCR6 Th1, Tc1, NK CXCL16 CC receptor CCR1 M, T, Eo, Ba CCL3, 5, 7, 14, 15, 16, 23 CCR2 alt M, Th1, DC, NK, Ba, Fib CCL2, 7, 8, 13
CCR3 Eo, Ba, Th2, P CCL2, 5, 7, 8, 11, 13, 15, 24, 26, 28 CCR4 M, Th2, DC, Ba CCL17, 22 CCR5 M, Th1, Tc1, NK CCL3, 4, 5, 7, 8, 11, 13 CCR6 alt T, B, DC CCL20 CCR7 T,DC CCL19, 21 CCR8 Th2, M, NK, DC CCL1 CCR9 alt T, IgA-Pla CCL25 CCR10 T CCL27, 28 C receptor XCR1 T, NK, P, Thy XCL1, 2 CX3C receptor CX3CR1 T, NK, M, En CX3CL1 Silent receptor CXCR7 CXCL111, CXCL12, MIF CCR11 alt CCL19, 21, 25 Duffy Ery, En CXCL1, 2, 3, 5, 6, 7, 8, 11 CCL2, 5, 7, 11, 13, 15, 17, 18, 22 D6 Placenta CCL2, 3, 4, 5, 7, 11, 13, 17, 22
alt, presence of alternative splicing(s)
Table 1.3 – Classification of all human chemokine receptors. B, lymphocyte B; Ba, basophil; DC, dendritic cell; Eo, eosinophil; En, endothelium; Ery, erythrocyt; Fib, fibroblast; IgA-Pla, IgA+ plasmacell; M, monocyt/macrophage; N, neutrophil; NK, natural killer cell; P, platelet T, lympho-cyte T; Tc1, type 1 killer T cell; Th1/2, type 1/2 helper T cell; Thy, thymolympho-cyte
on Chemotactic Cytokines” in 1996, divided into four families: CC chemokine receptors, CXC chemokine receptors, one C chemokine receptor and one CX3C chemokine receptor that correspond to the four distinct subfamilies of chemokines they bind – Table 1.3, p. 14. In addition, chemokine receptors with structural fea-tures that are inconsistent with a signaling function have been described. When ligated, these “silent” (i.e., non-signaling) chemokine receptors do not elicit migra-tion or convenmigra-tional signaling responses, but instead damped the immune response by binding, internalizing, and, in the case of D6, degrading chemokines – Table 1.3, p. 14. Chemokine decoy receptors recognize distinct and complementary sets of ligands and are strategically expressed in different cellular contexts. The abil-ity of chemokine receptors to signal upon ligand binding is due, at least in part,
to the presence of a DRY (glutamic acid/aspartic acid-arginine-tyrosine) motif in the second intracellular loop, which is missing in the scavenger receptors [58]. No chemokine receptor structures have been solved to date, and models are generally based on bovine rhodopsin, the only seven-transmembrane receptor for which three-dimensional structures have been solved.
Chemokine binding on leukocytes leads to the rapid up-regulation of adhesion molecules: integrins like very late antigen-4 (VLA-4) (or α4β1), which are needed to bind more strongly epithelial cells orintercellular adhesion molecule (ICAM)and vascular cell adhesion molecule (VCAM) (both overexpressed in response to proin-flammatory cytokines like IL-1, TNF-α or IFN-γ). Lastly, chemokine binding induced reorganization of the cytoskeletal, which finally leads to the cell polariza-tion, migration and infiltration of activated leukocytes.
1.4 Goals of this work
1.4.1 Epidemiology of gynecological tumors
Breast cancer is a heterogeneous disease embracing a range of biologic behaviors and prognostic characteristics [59]. Worldwide, breast cancer is the most common invasive cancer in women, representing 23% of all cancers diagnosed in women in 2008 [60]. The same year, breast cancer resulted in 1.38 million diagnosed cases and 458,000 deaths [60]. During the last decades, early diagnosis and multiple therapies helped to improve survival rate. Survival rates in the developed world are high, between 80% and 90% of the women being alive 5 years after the first diagnosis [59, 60]. But since the 1970s, the number of case has significantly increased, possibly due to modern living style, better diagnostic or longer lifetime. In addition, the situation in the developing countries remains problematic [61].
Among gynecological tumors, ovarian cancer has the highest mortality rate. Worldwide, over 200,000 women are diagnosed with primary ovarian cancer every year [62]. These cancers are mostly diagnosed at a late stage, over 75% of patients are already presenting metastasis at first diagnosis. Overall five-year survival rate ranges between devastating 25% and 49% [63]. Although standard therapy initially leads to good response rates, the disease recurs in over 50% of the cases within the following five years [64]. The rate of ovarian cancer between 1993 and 2008 de-creased in women of the 40-49 age cohort and in the 50-64 age cohort, possibly due to this group’s widespread adoption of oral contraceptives [62].
1.4.2 Current problematic
Current therapeutics often fail to eliminate all tumor cells and tumors with reduced immunogenicity may escape after a period of equilibrium due to a process called immunoediting [31, 65]. This concept holds that the immune system not only pro-tects the host against development of primary nonviral cancers but also sculpts
tu-mor immunogenicity [66]. Limitations of current therapies have led to increasing enthusiasm for defining new prognostic tools and highly targeted therapies.
1.4.3 Working hypothesis
During the last decade, three chemokine receptors (i.e., CCR4, CCR8, and CCR10) were found to be specifically expressed on Treg. With their respective associated chemokines, they all play a pivotal role in Treg migration. CCR4 and CCR8 were the first receptors to be described on Treg [67]. Together with CCL22, the ligand for CCR4, they lead to Treg migration into tumors as described in ovarian can-cer [15]. CXCR4 and CCR7 were also found expressed on Treg, although not spe-cifically [44, 45]. Bone marrow strongly expresses CXCL12, the ligand for CXCR4, leading to human Treg traffic and accumulation. The CCR7 pathway is very impor-tant for the Treg migration into lymph nodes and the acquisition of their full sup-pressive functions. More recently CCR10 was found overexpressed on Treg [16]. In a tumors hypoxia context, it was shown that CCL28 promotes tumor tolerance in a Treg dependent mechanism.
1.4.4 Purpose of this work
Although previous studies associated increased intratumoral number of CD4+CD25+ Treg with CCL22 and CCL28 and reported adversely affected prognosis in ovarian cancer, before our current work the clinical importance of other chemokines was unknown [15, 16]. Since several other chemokines were described to be potentially responsible for Treg accumulation in tumors and therefore, represent potential tar-gets for cancer therapy, we aimed to investigate whether other chemokines could lead to specific migration of Treg. The purpose of this work was to assess the clini-cal significance of different Treg attracting chemokine intratumoral expressions in breast cancer and ovarian cancer patients with long-term follow-up. Thus, we here investigated the prespecified hypothesis that the expression of Treg specific chemo-kines in cancer tissue predicts the OS and could therefore be used as biomarkers or as targets for immunotherapy.
* * *
—Anything found to be true of E. coli must also be true for elephants.
Jacques Monod, molecular biologist, 1954
2.1 Lymphocyte acquisition and manipulation
2.1.1 Peripheral blood mononuclear cell isolation
H
umandonor peripheral blood by Biocoll (peripheral blood mononuclear cells (PBMC)Merck, Darmstadt, Germany) density gra-were purified from healthy dient centrifugation. All samples were collected in the laboratory and were extempo-raneously performed. Briefly, peripheral blood was two times diluted in phosphate-buffered saline (PBS)(GE Healthcare, Munich, Germany) and 30 mL were then care-fully set down on 15 mL Biocoll solution in a 50 mL tube. Tubes were centrifuged at 1000 g, 4°C for 20 minutes, without centrifugal brake. The formed PBMC ring was quickly removed in order to minimize interaction with Biocoll and cells were washed three times in PBS.2.1.2 Lymphocyte maintenance and culture
Cells were then maintained inRoswell Park Memorial Institute medium (RPMI)1640 (Merck, Darmstadt, Germany) with 10% heat-inactivated fetal calf serum, 2 m mo-lar (M) L-glutamine, 1 mM sodium pyruvate, 1% nonessential amino acids, 100 international unit (IU)/mL penicillin, 100 mg/mL streptomycin and 0.05 mM β-mercaptoethanol supplemented with 10 ng/mL IL-2 (i.e., 180 IU/mL) (all from Life Technologies, Paisley, United Kingdom). PBMC were set up for migration assays as soon as possible and were never kept in RPMI more than 12 hours.
2.1.3 Migration assay
Chemotaxis assays were performed using 3 µm pore polycarbonate filters in a 96-well trans96-well chamber (Corning, New York, USA). Total PBMC (106 cells/well), after being washed three times with PBS, were added in migration medium (RPMI 1640 supplemented with 0.5% bovine serum albumin and 10 ng/mL IL-2) on the top chamber. Different concentrations of chemokines (all from Peprotech, Ham-burg, Germany) were added in the same medium supplemented on IL-2 to the lower well. All experiments were made at least in triplicate and migration occurred during three hours at 37°C and 5% CO2. Cells migrated into the lower chamber were then washed and labeled with antibodies before being analyzed by flow cytom-etry.
Total migrated cells were counted using CountBright absolute counting beads (Life Technologies, Paisley, United Kingdom). Migration results are shown as chemotaxis index (CI), which represents the total number of migrated cells in re-sponse to a chemokine divided by the total number of migrated cells without che-mokine. Theenrichment factor (EF) was obtained dividing the initial percentage of cells by the final percentage of cells and adding 1.
2.1.4 Flow cytometry analysis
PBMC migration was analyzed using the BD FACS Canto II cytometer (BD Bio-sciences, Heidelberg, Germany). Antibodies (anti-CD4 PacBlue, anti-CD25 APC and anti-CD127 PE, all fromBiolegend, San Diego, USA) were added to the dif-ferent cell populations and incubated for 20 minutes at 4°C in the dark, before being washed with PBS and analyzed. Gates were set according to side scatter and forward scatter on the lymphocyte population before gating on CD4+ CD25highCD127low cells – Figure 2.1, p. 19. CD4+ CD25high CD127low labeling shows 90% correla-tion with FoxP3 expression [35]. The percentage of positive cells of each individual marker were calculated usingFlowJo(FlowJo, Ashland, USA).
2.2 Chemokine receptor mRNA level expressions
2.2.1 Cell sorting
Isolation of the CD4+ CD25high CD127low population was performed in a two-steps procedure directly from the freshly isolated PBMC using a CD4+ CD25high CD127low Treg isolation kit (Miltenyi Biotec, Bergisch Gladbach, Germany). The non-CD4+CD127highcells were first magnetically labeled with a cocktail of biotin-conjugated antibodies and secondary anti-biotin monoclonal antibodies. Labeled cells were depleted from the total population by separation over aMACS LDcolumn placed in the magnetic field of aMACSseparator.
During the second step, the CD4+ CD25high CD127low Treg were directly la-beled among the CD4+CD127lowenriched population with CD25highmicro-beads
Figure 2.1 – Gates were set according to side scatter and forward scatter on the lymphocyte
popu-lation before gating on CD4+CD25highCD127lowcells (red square).
and isolated by positive selection from the pre-enriched fraction by separation over aMACS MScolumn placed in the magnetic field of aMACS separator. During the separation, cells were kept cold, and all solutions were pre-cooled in order to prevent capping of antibodies on the cell surface and non-specific labeling. Cells were re-suspended in a buffer made from PBS, 0,5%bovine serum albumin (BSA)and 2mM ethylenediaminetetraacetic acid (EDTA). Antibodies were added as recommended by the fabricant and incubated 15 minutes at 4°C in the dark. Cells were three times washed before magnetic separation and then rapidly further performed.
2.2.2 mRNA isolation
After isolation cells were washed three times with pre-cooled PBS. Total cellular ribonucleic acid (RNA) was extracted from the different cell populations using the QuickGene RNA blood cell kit S and theQuickGene mini 80automat (Fuji, Tokyo, Japan). Briefly, leukocyte pellets were pooled down in 1.5 mL tubes, resuspended in 520 µL of LRB(after reconstitution with adequate volume of β-mercaptoethanol) and thoroughly mixed by vortexing for 30 secondes. Finally, 250 µL of 99% ethanol
Receptor Forward Reverse Sonde CCR1 5’-TCTGACTCTTGGCACAGCAT-3’ 5’-GCCACCATTACATTCCCTCTC-3’ #76 CCR2 5’-TGAGACAAGCCACAAGCTGA-3’ 5’-TTCTGATAAACCGAGAACGAGAT-3’ #56 CCR3 5’-TGCAGTGCTCTTTACCCAGA-3’ 5’-GGTCATTCTCAGAGTGTGGAAA-3’ #78 CCR4 5’-TTCGTGTTTTCCCTCCCTTT-3’ 5’-ACCTAGCCCAAAAACCCACT-3’ #84 CCR5 5’-AACCAGGCGAGAGACTTCTCA-3’ 5’-GATCCAACTCAAATTCCTTCTCA-3’ #14 CCR6 5’-AAAGGCATCTATGCCATCAACT-3’ 5’-GTACCGGTCCATGCTAATGC-3’ #68 CCR7 5’-GGGGAAACCAATGAAAAGC-3’ 5’-ACCTCATCTTGACACAGGCATA-3’ #77 CCR8 5’-TGCCTCCTGTTTGTATTCAGTCT-3’ 5’-CAGACCACAAGGACCAGGAT-3’ #64 CCR9 5’-CACCATGACACCCACAGACT-3’ 5’-TCACAGTAGAAGTCAGTGAAGTTGAA-3’ #21 CCR10 5’-AGTAGGTGGGGGAACACTGA-3’ 5’-GCAAGGCACAGAGGTAGTCC-3’ #34 CCR11 5’-TGCAAATGAGAAAAACACTAAGGT-3’ 5’-TGGCAAAAACAAGCTTGAAA-3’ #32
Table 2.1 – Primers used for the qPCR analysis. All primers were designed using theRoche assay design center– #number corresponds to probe from theUniversal Probe Library.
was added to each tube. Tubes were thoroughly mixed by vortexing for 30 secondes and kept at 25°C for 5 minutes. After a flash spin-down, whole lysate were trans-ferred to the cartridge and performed for one minute under pressure. Cartridges were washed three times with 750 µL WRBsolution (after reconstitution with ad-equate volume of ethanol). Elution was performed in RNAse-free water for one minute under pressure after waiting for 30 secondes at 25°C. Total RNA concentra-tion and purity was determined with aNanodrop(Thermo Fisher Scientific, Waltham, USA). 260 nm absorption was compared with 280 nm absorption and the ratio op-tical density (OD)260 nm/OD280 nm was routinely ≥2. If not immediately used, total RNA was stored at -20°C.
2.2.3 cDNA transcription
complementary DNA (cDNA)from total mRNA was synthetized using theFermentas RT-kitand oligo(dT) primers (Fermentas, Waltham, USA). Briefly, up to 1 µg of total RNA was mixed with 2 µL 10X reaction buffer, 1 µL oligo(dT) primers, 2 µL 10 nM dNTP mix and completed to 20 µL with nuclease-free water in polymerase chain reaction (PCR) tubes. Oligo(dT) primes cDNA synthesis from the poly(A) tail present at the 3’-end of eukaryotic mRNA and thus allow to obtain a copy from all mRNA previously isolated. PCR tubes were placed in a thermal cycler and warmed at 37°C for 30 minutes to allow the transcription and 10 minutes at 72°C to terminate every reaction before cooling down to 4°C. Total cDNA concentration and purity was determined with a Nanodrop (Thermo Fisher Scientific, Waltham, USA). 260 nm absorption was compared with 280 nm absorption and the ratio OD260 nm/OD280 nm was routinely≥1.8. If not immediately used, total cDNA was stored at -20°C.
2.2.4 RT-qPCR analysis
The following synthetic primers (allEurofins Scientific, Luxembourg, Luxembourg) were used for thereal-time PCR (qPCR)(designed using theRoche assay design center – #number correspond to probe from theUniversal ProbeLibrary) – Table 2.1, p. 20.
using theLightCylcer 480(Roche, Basel, Switzerland). Reaction were performed in the following mix (per well): water 3 µL, sonde 0.2 µL, primers 0.4 µL + 0.4 µL (forward and reverse), master mix 5 µL (VWR International, Radnor, USA) and cDNA 2 µL. Quantity values generated for gene expression were obtained by com-parison of the fluorescence generated by each sample with hypoxanthine-guanine phosphoribosyltransferase (HPRT) standard curves. As control of Treg purity, the level of FoxP3 and CD127 mRNA was quantified in isolated cell populations. As control of RNA isolation purity, qPCR were performed with 2 µL from isolated RNA solutions.
2.3 Histopathology
2.3.1 Breast cancer patient recruitment
Microarrayed tissue samples from breast cancers were used in this retrospective study. Samples were collected from patients undergoing surgery at the Gynecology unit (Klinik und Poliklinik für Frauenheilkunde und Geburtshilfe, Ludwig-Maximilians-Universität, Munich).
All cases had been previously characterized in terms of histology (nodal status, tumor size, grade,estrogen receptor (ER)-α/progesteron receptor (PgR)-α/ human epi-dermal growth factor receptor 2 (Her2)status at the Instituty of Pathology, Ludwig-Maximilians-Universität, Munich. The histological classification was determined according to a modification of the Elston and Ellis grading proposed by Bloom and Richardson (Elston and Ellis, 1991). The hormone receptor status was evaluated by immunohistochemistry. The tumor was classified as hormone receptor positive in case of positive staining in ≥ 10% of the tumor. Normal breast tissue was ob-tained from women undergoing breast reduction surgery (n= 7). Criteria for the selection of patients for the tissue microarray were adequate routinely fixed biopsy material, no history of previous malignancy but history of distant metastases during the course of the disease.
A total of 180 ductal tumors, 14 lobular tumors and 5 others were used for this study. No preselection of patients was made. All available patients suitable for the inclusion were selected. Patients were not matched for stage disease or age. Patients were diagnosed from 1986 to 2007 and almost all underwent surgery within one month (max 7 months). All patients received standards surgical treatment of either mastectomy or wide local excision with radiotherapy. Systemic adjuvant treatments were given based on clinical scores and hormone receptor status. The end of the follow-up period was October 2014. Additional patient information is listed in – Table 3.3, 3.4, and 3.5, pp. 37, 38, 39. Of the 199 tumors on the array, 192 presented an assessable FoxP3 staining, 193 an assessable CCL1 staining, 194 an assessable CCL22 straining, and 192 an assessable CCL27 staining.
All materials was sampled for diagnostic purposes and research was done in accordance with the legal requirements concerning confidential medical
commu-nication as well as the data protection act. Consequently, consulting the ethics committee of the medical school, Ludwig-Maximilians-Universität, Munich, and written informed consent from the patients prior to participation in the study was not required.
2.3.2 Ovarian cancer patient recruitment
203 tumor tissue samples were selected from the archives of the Institute of Pathol-ogy of the Ludwig-Maximilians-Universität, Munich for this study. All patients had undergone debulking surgery at the Klinik und Poliklinik für Frauenheilkunde und Geburtshilfe, Munich, and had been previously characterized in terms of histol-ogy at the Institute of Patholhistol-ogy of the Ludwig-Maximilian-Universität, Munich.
The patients were between 27 and 87 years of age at diagnosis, with a median age of 60. tumors were graded according to the Silverberg grading system. Most tumors were graded class 3+4 (62.2%) and only very few class 1 (2.96%). Only samples classified as stage III cancers, according to theInternational Federation of Gynecology and Obstetrics (FIGO), were included. All histological types of epithelial ovarian carcinoma were selected, but the serous (77.1%), endometroid (8.1%) and adenocarcinoma NOS (5.3%) were by far the most commonly represented types. Selection criteria of patients for the inclusion in this study were previous inclusion and remain of adequate routinely fixed biopsy material (203 of the total 210 orig-inally included patients) [68]. Additional patient information is listed in – Table 3.1 and 3.2, pp. 33, 34. Of the 203 tumors, 203 presented an assessable CCL1 staining.
All materials was sampled for diagnostic purposes and research was done in accordance with the legal requirements concerning confidential medical commu-nication as well as the data protection act. Consequently, consulting the ethics committee of the medical school, Ludwig-Maximilians-Universität, Munich, and written informed consent from the patients prior to participation in the study was not required.
2.3.3 Immunohistochemistry
Paraffin wax-embedded tissue slide samples were deparaffinized in xylol for 20 min-utes, washed in 100% ethanol and then incubated in methanol/H2O2 (3%) for 20 minutes. After rehydration in an alcohol gradient to distilled water, the slides were placed in a pressure cooker containing sodium citrate buffer (pH= 6.0) and cooked for 5 minutes. Slides were washed twice in PBS and blocked using blocking solution 1 fromZytoChem Plus horseradish peroxidase (HRP)Polymer Kit (Zytomed, Berlin, Germany) for 5 minutes. Each slide was separately incubated with a polyclonal rabbit anti-human CCL1 primary antibody (Atlas antibodies, Stockholm, Sweden) diluted 1/200 in PBS, a polyclonal rabbit anti-human CCL22 primary antibody (Peprotech, Hamburg, Germany) diluted 1/200 in PBS, a polyclonal rabbit anti-human CCL27 primary antibody (Atlas antibodies, Stockholm, Sweden) diluted
1/200 in PBS or a monoclonal mouse anti-human FoxP3 primary antibody (clone 236A/E7) (Abcam, Cambridge, USA) diluted 1/300 in PBS. Incubation of the sec-tions with the primary antibodies lasted for 16 h at 4°C. Afterward, secsec-tions were washed twice in PBS before incubation with postblock reagent 2 (Zytomed, Berlin, Germany) for 20 minutes. The slides were then washed in PBS and then incubated with the HRP-polymer 3, containing secondary antibody coupled with detection enzyme (Zytomed, Berlin, Germany) for 30 minutes. Staining was performed us-ing3,3-diaminobenzidine (DAB)-substrate solution (Dako, Glostrup, Denmark) for 180 seconds. Counterstaining was carried out with Mayer’s hemalaun for 2 minutes. Finally, sections were washed in tap water for 5 minutes and afterwards dehydrated in an ascending alcohol serie and washed in xylol. Slides were cover-slipped with Eukittquick-hardening mounting medium (Sigma-Aldrich, St. Louis, USA).
2.3.4 Quantification of immunohistochemistry
For each labelling, absolute numbers of positive cells were computer-aided deter-mined. Without any knowledge of identity, pictures were systematically taken (ar-rays contained duplicate cores) and positive cells were automatically counted using theimage Jsoftware (Version 1.49o, National Institutes of Health, Bethesda, USA). DAB labelling was extracted using the IHC Toolboxplugin and cells were then de-fined and counted based on their size and morphology – Figure 2.2, p. 24.
2.4 Statistical analysis
2.4.1 Statistical program
The statistical programR(Version 0.99.441,RStudio, Inc., Boston, USA) was used for data collection and processing as well as analysis of statistical data. Following packages were used:
• U. Ligges and M. Mächler (2003). Scatterplot3d - anRPackage for Visual-izing Multivariate Data. Journal of Statistical Software 8(11), 1-20.
http://www.jstatsoft.org.
• RCore Team (2015). R: A language and environment for statistical comput-ing. RFoundation for Statistical Computing, Vienna, Austria.
http://www.R-project.org/.
• T. Therneau (2015). A Package for Survival Analysis inS. version 2.38.
http://CRAN.R-project.org/package=survival.
• H. Wickham (2009). ggplot2: elegant graphics for data analysis.
Figure 2.2 – Representative examples of computer-aided determined labelings. (a-b) Whole
breast cancer tissue section with CCL1 recognition. (c-d) Whole breast cancer tissue section with CCL22 recognition. (e-f ) Whole breast cancer tissue section with CCL27 recognition. (g-h) Whole breast cancer tissue section with FoxP3 recognition. All magnifications× 200.
2.4.2 Statistical tests
The Kruskal-Wallis test was used to compare more than two independent groups, and the Mann-Whitney U-test was used for evaluation of two independent groups. These tests are one-way analysis of variance and analyze two or more samples which are independent from each other. Contingency tables were analyzed using the Pear-son’s χ2 test. Linear regression were realize using the Pearson’s product-moment correlation or Spearman’s Rsbased on the continuity of analyzed values. The choice betweeninterquartile range (IQR) orstandard error of the mean (SEM) is detailed in the relevant figure legends. Patients included in the statistical models are those for which all the necessary data was available. A small number of patients did not have complete data. Values with p≤ 0.05 were considered statistically significant.
2.4.3 Survival analysis
Given that there is no clinical defined cutoff point for the number of chemokine-expressing cells in such a context, we always selected a cutoff at the median values because this divided the patients into equal-sized groups, and does not make the assumption of an artificial cutoff for statistical analyses. We systematically analysed whether there was any correlation between cell numbers and age, nodal status, tu-mor size, tutu-mor rest, grade, ER-α/PgR-α/Her2 status, adjuvant therapy, and OS. Survival was measured from the date of diagnosis to the time of death or the time the patients was last seen. The log-rank test was used to perform univariate analyses and the survival curves were estimated by theKaplan-Meier (KM)method. Prognos-tic factors for survival were evaluated in multivariate analyses by Cox proportional hazards regression. The statistical tests are detailed in the relevant figure legends.
2.4.4 Kaplan-Meier curves with ggplot2
Unfortunately, the ggplot2 package does not accept asurvfitobject for
representa-tion of KM curves. In order to plot the curves with this package, following code was modified and used in this work:
##################################### ##### P l o t t i n g KM u s i n g ggplo t2 ##### ##################################### # Define f u n c t i o n
ggkm <− function( s f i t , r e t u r n s = FALSE,
x l a b s = ”Time a f t e r Surgery ( Months ) ”, y l a b s = ” Proportion S u r v i v i n g ”,
y s t r a t a l a b s = NULL, y s t r a t a n a m e = NULL, timeby = 50 , main = ” O v e r a l l S u r v i v a l ”, p v a l = TRUE, . . . ) {
